The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill downloads configuration metadata and data CSVs from the github.com/SalesforceLabs/LSStarterConfig.git repository. This is a well-known repository owned by the vendor (Salesforce), and the download is used to provision the target environment.
[COMMAND_EXECUTION]: The skill utilizes the Salesforce CLI (sf), git, and jq to perform organizational configuration, metadata deployment, and data manipulation. These are standard tools for Salesforce development and administration.
[DYNAMIC_EXECUTION]: The skill generates and executes temporary anonymous Apex scripts (e.g., .lsc-setpw.apex) via the sf apex run command. These scripts are used for tasks such as setting user passwords and managing metadata cache records. The scripts are constructed from predefined templates within the skill's reference files.
[INDIRECT_PROMPT_INJECTION]: The skill ingests data from external CSV files to create and update Salesforce records.
Ingestion points: Data is read from CSV files located in the .lsc-starter-config/LSStarterConfig/Data/ directory after being downloaded from a trusted source.
Boundary markers: No explicit boundary markers are used for the CSV content during interpolation into CLI commands.
Capability inventory: The skill has the capability to create/update records and run Apex code.
Sanitization: No specific sanitization of CSV content is mentioned; however, the data is used for structured record creation which limits the impact of potential injection.
[PROMPT_INJECTION]: The skill contains instructions to perform "silent" cleanup of temporary directories and files (e.g., Phase 8 in SKILL.md). This is a standard operational practice in automation to maintain workspace cleanliness and protect sensitive temporary data, rather than an attempt to bypass safety guidelines.

life-sciences-fieldsalesrep-coordinate